The application of micro computed tomography
to assess quality of parts manufactured by means
of rapid prototyping
Bartosz Gapiński1), Michał Wieczorowski1), *), Mirosław Grzelka1), Pedro Arroyo Alonso2), Adrián Bermúdez Tomé2)
DOI: dx.doi.org/10.14314/polimery.2017.053
Abstract: In the paper a possibility of using a metrology micro computed tomography (CT) to assess
quality of parts made by rapid prototyping technique was presented. Additive manufacturing tech-niques dramatically change the approach to manufacturing process. Not long ago they allowed solely for obtaining a spatial model of a part in order to visualize its features. Nowadays, it is possible to produce elements and assembly groups with properties that make it possible to replace many parts manufactured using traditional casting or injection techniques. Polymer components created by means of 3D printing can be generated one layer of material after another. In many cases inner spaces (sec-tions) are not fully filled in (solid), they are rather replaced by a specially selected structure. Thanks to this it is possible to save material and reduce weight. To control these structures and to inspect joints between particular layers using a non-destructive method it is necessary to use a metrology micro CT. In the paper an application of X-ray computed tomography and assessment of measurement uncertainty regarding elements manufactured with fused deposition modeling (FDM) method using ABS 3D print-ing on UP! printer was presented. Inspection of both: inner structure and outer shapes as well as joints between them was performed basing on computer aided design (CAD) model.
Keywords: computed tomography, uncertainty of measurements, rapid prototyping.
Zastosowanie tomografii komputerowej do oceny jakości elementów
wykonanych metodą szybkiego prototypowania
Streszczenie: W pracy przedstawiono możliwość zastosowania pomiarowej tomografii komputerowej
(CT) do oceny jakości elementów wykonanych techniką szybkiego prototypowania. Techniki wytwa-rzania addytywnego rewolucjonizują podejście do technik wytwawytwa-rzania. Jeszcze niedawno pozwalały jedynie na uzyskanie przestrzennego modelu w celu jego wizualizacji. Obecnie możliwe jest uzyskanie gotowych elementów o właściwościach pozwalających na zastąpienie wielu wyrobów wykonywanych tradycyjnymi technikami odlewania lub wtryskiwania. Elementy polimerowe uzyskiwane za pomo-cą druku 3D wytwarzane są przez nakładanie kolejnych warstw materiału. W wielu wypadkach we-wnętrzne przestrzenie (przekroje) elementu nie są całkowicie wypełnione, ale są zastępowane przez odpowiednio dobraną strukturę wewnętrzną. Pozwala to na oszczędność materiału i redukcję masy elementów. Kontrola jakości tych struktur, jak i połączeń pomiędzy poszczególnymi warstwami, jest możliwa w sposób nieniszczący jedynie za pomocą pomiarowej tomografii komputerowej. Metodę rent-genowskiej tomografii komputerowej zastosowano do oceny niepewności pomiaru elementów wykona-nych z ABS techniką osadzania topionego materiału (FDM) z zastosowaniem drukarki UP! 3D. Doko-nano badania i oceny na podstawie modelu projektowanego ze wspomaganiem komputerowym (CAD) z uwzględnieniem zarówno powierzchni zewnętrznych, jak i wewnętrznej struktury oraz połączeń pomiędzy poszczególnymi elementami.
Słowa kluczowe: tomografia komputerowa, niepewność pomiarów, szybkie prototypowanie.
1) Poznan University of Technology, Institute of Mechanical Technology, Division of Metrology and Measurement Systems,
Piotrowo 3, 60-965 Poznań, Poland.
2) University of Vigo, Faculty of Industrial Engineering, Rúa Conde de Torrecedeira 86, 36208 Vigo, Pontevedra, Spain.
Coordinate measuring technique has been continual-ly developed for about fifty years. It comprises different techniques and measurement devices, operating in 2D and 3D. A two-dimensional measurement is possible for example on a measuring microscope or optical machines. The most widely used are instruments that enable to in-spect parts in three dimensions (space). Here, the best known are coordinate measuring machines (CMM) and measuring arms [1, 2]. For about 15 recent years optical devices are more and more commonly applied (e.g., scan-ners using different light patterns). All these instruments have a common feature – they are suitable for measuring elements, where it is possible to make contact with a sur-face (tactile methods) or to observe it (contactless meth-od). Still, they do not allow to inspect geometry of inside – invisible elements, without destroying the part [3–5]. An alternative device to that is an X-ray tomograph. In medicine they are known from 70-ties of XX century, but in mechanical engineering applications they are used for last couple of years. Apart from surface that can be mea-sured with more traditional instruments they allow to in-spect closed spaces as well as assess quality of assembly and functionality of groups in ready mechanisms [4, 6].
COMPUTED TOMOGRAPHY
An expression tomography comes from Greek words “tomé” (section) and “gráfein” (write). As a combination of projections of a measured object executed from differ-ent directions it is possible to create section images (2D) and spatial images (3D). Analysis and working out of par-ticular projections is performed by means of a computer – hence the name computed tomography (CT). Mathe-matical bases of this invention were made by an Austrian mathematician Johann Radon, who in 1917 proved that an image of two- and three-dimensional object can be reconstructed in a total from infinite number of projec-tions of the same part [7].
Metrology tomography is a kind of X-ray tomography. The first computed tomograph enabling to obtaining an image of inner structure was elaborated between 1969 and 1972 by Godfrey Newbold Hounsfield. Its production version was released to the market in 1973 and – due to a company that sponsored the whole project – is known as EMI Scanner. Hounsfield was given a Nobel Prize in medical sciences for his achievements in 1979 [8–10].
A basic way of inspection is to focus an X-ray beam onto an object and registering its intensity on a detector on the other side. X-ray beam, crossing an inspected object, loses its power, and this weakening is a function of radiation energy as well as the kind and thickness of the inspected material. The volume of material is divided into small el-ements, called voxels, in which linear coefficient of radia-tion absorpradia-tion is the same. A reconstructed secradia-tion im-age is a quantitative map of a linear radiation absorption coefficient in voxels, being a part of a scanned layer. Cre-ation of a tomographic image is illustrated in Fig. 1.
An X-ray radiation source in measurement tomographs is an X-ray lamp. The first commercial X-ray lamp was constructed in 1913 by W.D. Coolidge. He used a heating of a tungsten fiber to form an electron beam in a glass pipe with vacuum. In contemporary systems two basic types of lamps are used: transmission and directional. Both types are presented in Fig. 2. A transmission lamp enables to reach a higher magnification, while a directional lamp is used when a greater power is needed [7, 12, 13].
In order to obtain better resolution of a measured im-age two kinds of lamps are used: microfocus and nano-focus. Particularly the latter one, additionally equipped with a special pinhole and voltage stabilization, makes it possible to reduce a spot in focus to a very small size, even below 1 micrometer [14].
UNCERTAINTY OF MEASUREMENT
Uncertainty of measurement is defined as a non-nega-tive parameter attributed to the measured quantity that is characterizing dispersion of measured values. Thus
un-a)
d) c)
b)
1
projection projections2 projections4 projections8 projectionsAll 0° + 45° + 90° + 135° 0° + 45° + 90°
0° + 45° 0°
Fig. 1. The idea of a tomographical image: a) measured object, b) X-ray radiation beam for a single measurement projection, c) a rule of connecting results of single projections, d) a result of a reconstruction of a real object measurement for various num-bers of projections (according to [11])
a) b)
Fig. 2. Types of lamps used in metrology tomographs: a) trans-mission, b) directional (according to [14])
certainty reflects incomplete knowledge of the measured quantity [15–19]. A proper evaluation of uncertainty is a very complex and time consuming process, because of a large number of sources and factors influencing its value. Generally, there are two methods to evaluate measurement uncertainty. Type A is based on statistical analysis while type B is used when applying statistics is not possible or justified. In type A, it is often assumed that a distribution best describing an input quantity is a Gaussian one, or – when a number of measurements is small – a Student’s t-distribution. Depending on the phe-nomenon other distributions apply as well.
For a type B evaluation, regardless of the lack of statis-tical knowledge, it is also possible to use a distribution. As the only information we have is that a value lies with-in a specified with-interval, a rectangular probability distribu-tion can be used, characterized by interval limits. If there is more information available, another probability distri-bution that comply with that information can be used.
In description of measurement uncertainty evaluation of a standard uncertainty u was introduced, expressed by means of standard deviation, as well as extended un-certainty U, considered as a quantity defining an interval dependent on assumed confidence level and expressed by a formula:
U = k · u (1)
where: U – extended uncertainty, u – standard uncer-tainty, k – extension coefficient, dependent on assumed confidence level p, where in practice it is assumed that
k = 2, what corresponds to confidence level p = 95 %.
Evaluation of type A uncertainty is based on a series of repeated measurements of certain quantity what is done by experiment. Standard deviation sx is considered as a standard uncertainty u: n )2 x xi ( sx ) x ( u n 1 1 − − = ≈
∑
(2)where: u(x) – standard uncertainty of a measurement,
sx – standard deviation, xi – value in a data set, x – arith-metical mean value from a data set, n – number of ele-ments in a set.
Standard deviation of sample distribution based on measured values sx is an assessment of a standard devia-tion regarding distribudevia-tion σ. Mean value x from a cer-tain number n of measurement results xi is an assessment of a correct mean value of the distribution.
Reducing the obtained value of standard uncertainty is possible by means of calculating x value from a series of measurements. In that case measurement uncertainty is n times smaller than standard uncertainty of a single measurement: ) n ( n )2 x xi ( n ) x ( u ) x ( u n 1 1 − − = =
∑
(3)where: u(x) – uncertainty obtained from calculation of the mean value from a series of measurements.
In industry reality because of a necessity to reduce measurement time it is very seldom possible to have enough time to do a large number of measurements in order to obtain a reasonable mean value. Another issue is to maintain constant measurement conditions and re-peatability of the process.
MICRO CT USED FOR RESEARCH
For research purposes a metrology micro CT was used. GE Phoenix v|tome|x s240 device, shown in Fig. 3, is equipped with two X-ray lamps. A directional lamp with large power enables working with voltages up to 240 kV and obtaining X-ray beam power up to 320 W and maximum magnification of the measured object up to 100 times. The second lamp is a transmission nanofocus one. It works with much less power, maximum 15 W, but its construction makes it possible to obtain much smaller focal spot size and magnifications up to 200 times. Thanks to this lamp it is possible to detect details as little as 0.5 µm. The measurement space of a CT enables to measure parts, with diameter up to 260 mm and 420 mm in length. Still, it is crucial to remember that the greater the density of a measured material, the smaller is a total thickness of walls that can be penetrated by X-ray radiation.
CREATION OF A REAL MODEL BY MEANS OF A MEM ADDITIVE MANUFACTURING
TECHNIQUE
A 3D model of a piston engine is presented in Fig. 4. The real model composed of: engine block, piston, crankshaft, and a pin was made using fused deposition modeling (FDM) method on an UP! 3D printer, shown in Fig. 5. A working space of a 3D printer was equal to 140 x 140 x 135 mm, what made it possible to print all
el-Fig. 3. Micro CT equipment type GE Phoenix v|tome|x s240 used for research
ements in 1:1 scale. A material used in the process was acrylonitrile-butadiene-styrene (ABS) terpolymer, having properties consistent with those listed in Wholers Report from March, 2011 [20].
In order to realize a printing process of elements it was necessary to convert all the files to a format that could be supported by a rapid prototyping (RP) device. In most cases the models have to be saved as STL files. A file con-version process is based on describing shape geometry by means of mesh of triangles with respectively accurate degree of density.
The first step to start the printing process was retriev-ing an in advance prepared STL file usretriev-ing printer soft-ware and placing models in 3D space of a printer. Because of a complex geometry of printed elements, printing reso-lution in Z axis equal to 0.15 mm was selected. It makes possible to obtain possibly the most accurate surface re-construction. A maximum filling degree was also se-lected. Apart from basic layer model parameters, the parameters of supports (support layers) were defined. Depending on the shape of model and its location on a working platform, the amount of supporting material used to build the whole element varies. Complex geome-tries often enforce using large number of supporting
lay-ers. This is a factor making processing of the whole ele-ment more difficult.
After finishing of 3D printing obtained real models were dismounted and redundant supporting material was removed.
Connecting computer aided design/computer aided en-gineering (CAD/CAE) systems with RP methods makes it possible to shorten time from elaboration of a project of a device to its physical creation [21].
RESEARCH AND DISCUSSION
ASSESSMENT WITH CT METHOD OF QUALITY OF MANUFACTURED ELEMENTS
For research a simplified model of a reciprocating en-gine by means of 3D printing was made. It consists of an engine block, piston, connecting rod, crankshaft, and pin. Each of these elements was measured separately in order to confirm its correct manufacturing [22, 23].
An example of measurement of a piston is shown in Fig. 6. The obtained values were presented in a form of a color map of dimensional deviations of a real part from a CAD model. Measurement results analysis was
per-Fig. 4. A simplified model of a reciprocating engine: a) CAD mo-del, b) element manufactured by means of a 3D printing
a) b)
Fig. 5. Part on the printer UP! 3D used in this study
Deviation, nm +0.5 +0.4 +0.3 +0.1 +0.2 0.0 -0.1 -0.2 -0.3 -0.4 -0.5 Deviation, nm +0.5 +0.4 +0.3 +0.1 +0.2 0.0 -0.1 -0.2 -0.3 -0.4 -0.5
Fig. 6. Color map of dimensional deviations of reciprocating engine made of ABS by means of 3D printing in comparison to a CAD model: a) top view, b) bottom view
b) a)
formed using Volume Graphics Studio MAX software. It enables to perform geometrical analysis of inspected element, in comparison with CAD model, as well as de-tection of pores and analysis of wall thickness [22–24].
The outer shape of the part (shown in Fig. 5) was mea-sured on a CT. Such measurements naturally can be made for a number of various devices. Still only CT makes it possible to assess the inside structure of the element without destroying it. A section of an engine block is
presented in Fig. 7. The inner structure of the element is clearly visible as a net filling the inside of the body. Thanks to such images full inside geometry analysis is possible including all the reinforcements, their thickness as well as continuity across the whole section of the ele-ment – in all directions.
In the upper part of the engine block some dissections are present. They can be seen with a naked eye and also in the photo shown in Fig. 8. Still only CT measurement gives possibility to assess the magnitude of that crack and its influence on the inner structure of the part.
Defects in part were presented in Fig. 9. These cracks and discontinuities are possible to detect thanks to some options of measured data analysis. On 2D section one can clearly see how deep this dissection penetrates into the inspected element. It allows a user to decide whether the part can be accepted for use or not. Additionally it makes possible to introduce modifications in the tech-nological process to eliminate such defects in following elements.
Computed tomography makes also possible to assess quality of assembly, location, and cooperation of particu-lar elements in a ready mechanism. A section of inspect-ed engine model in the lowest position of the piston was shown in Fig. 10. Using volume graphics software it is possible to prepare a 3D view and 2D sections.
Computed tomography enables to measure a mecha-nism in several places, important for its functionality. In that case it is possible to assess geometrical changes and deformations of each single element.
MEASUREMENT UNCERTAINTY ANALYSIS ON CT
Measurement uncertainty is one of the quantities that helps to find out whether the results are useful or not. In this case the uncertainty was assessed for characteristic features of a piston. Results obtained as an effect of that analysis were shown in Table 1. Uncertainty was evalu-ated according to type A and the expansion coefficient
Fig. 7. A section through an engine block showing a structure inside the element
Fig. 8. A view of an engine block showing dissection
Fig. 9. A view (a) and section (b) of an engine block with visible dissection penetration inside the element
b) a)
T a b l e 1. Uncertainty results for characteristic features of piston
Element Measured Nominal dimensionmm Mean valuemm uncertainty UMeasurement
95, mm
Outer cylinder diameter 49.650 49.631 0.0113
cylindricity deviation 0 0.203 0.0015
Hole for pin diameter 10.000 9.744 0.0074
Inner cylinder diameter 35.700 35.682 0.0170
runout 0 0.016 0.0079
Front surface flatness 0 0.200 0.0042
distance from pin axis 17.000 16.945 0.0151
was assumed as k = 2. Analyzing the results presented in the table one can observe that the measurement uncer-tainty U95 ≤ 0.017 mm. This value shows that a computer tomograph is an accurate measurement system, bearing in mind, that among most of the inspected features the measurement uncertainty does not exceed 0.008 mm.
CONCLUSIONS
Metrology computer tomographs are the most mod-ern devices used in contemporary coordinate measuring technique. They allow to conduct research in a very wide range of applications, including inspection of part geom-etry (dimensions) and checking its inner structure. Data obtained as a result of measurement can be also analyzed from pores and discontinuities point of view, as well as to view changes of wall thickness.
A measurement performed with a CT method is a con-tactless one, what makes it possible to inspect surfaces with any shape. Still, it does not have drawbacks of opti-cal systems where it is often not possible to check narrow gaps or holes. The greatest limitation of that technique is possibility of penetration of measured object, depen-dent on the material density and total wall thickness in the examined object. As far as plastic elements are con-cerned this is very seldom the case limiting inspection possibilities. However, it is important to remember that as the power of lamp increases, the measurement accu-racy decreases. It is connected with a size of a focal spot.
At present CT method is not very popular, particularly in industry applications. One of the reasons is the fact that the technique is very young, another issue is high price of such devices. There are however opinions that during next 10–15 years even 20 % of dimensional mea-surements will be performed with CT methods, as in-spection possibilities look very promising.
REFERENCES
[1] “The Metrology Handbook, Second Edition” (Ed. Bu-cher L.J.), ASQ 2012.
[2] Howarth P., Redgrave F.: “Metrology – in short 3rd edition”, EURAMET project 1011, 2008.
[3] Kruth J.-P., Bartscher M., Carmignato S. et al.: CIRP
Annals – Manufacturing Technology 2011, 60, No. 2, 821.
http://dx.doi.org/10.1016/j.cirp.2011.05.006
[4] Gapinski B., Zachwiej I., Kolodziej A.: “Comparison of different coordinate measuring devices for part geometry control”, Proceedings of Digital Industrial
Fig. 10. An assembly model of the engine block: a) 3D view, b) 2D section along the cylinder axis
b) a)
Radiology and Computed Tomography Conference, Ghent, Belgium, June 22–25, 2015.
[5] Vanhove Y., Wang M., Wieczorowski M., Mathia T.G.: “Metrological perspectives of tomography in civil engineering”, Proceedings of 6th World Congress on Industrial Process Tomography, Beijing, China, September 6–9, 2010.
[6] Gapinski B., Wieczorowski M., Marciniak-Podsadna L. et al.: Procedia Engineering 2014, 69, 255.
http://dx.doi.org/10.1016/j.proeng.2014.02.230
[7] Wieczorowski M., Rozanski L., Grzelka M., Mathia T.: “Principles of computer tomography application in geometrical metrology”, Proceedings of Manu-facturing Conference, Poznan, Poland, November 24–26, 2010, p. 203.
[8] Maizlin Z.V., Vos P.M.: Journal of Computer Assisted
Tomography 2012, 36, No. 2, 161.
http://dx.doi.org/10.1097/RCT.0b013e318249416f [9] Oransky I.: The Lancet 2004, 364, No. 9439, 1032. http://dx.doi.org/10.1016/S0140-6736(04)17049-9 [10] Littleton J.T.: “Conventional Tomography” in “A
His-tory of the Radiological Sciences”, American Roent-gen Ray Society, (PDF file, retrieved 11 January 2014). [11] Christoph R., Neumann H.J.: “X-ray Tomography in Industrial Metrology, Precise, Economical and Uni-versal”, Verlag Moderne Industrie 2011.
[12] Ratajczyk E.: Mechanik 2011, 5–6, 474.
[13] Gapiński B., Marciniak-Podsadna L.: Mechanik 2014,
8–9, 59.
[14] www.ge-mcs.com (access date 2015.08.15).
[15] “International vocabulary of metrology – Basic and general concepts and associated terms (VIM)”, 3rd edition – JCGM 2012.
[16] Guide to the expression of uncertainty in measure-ment – JCGM 2008.
[17] Dewulf W., Kiekens K., Ye T. et al.: CIRP Annals –
Manufacturing Technology 2013, 62, No. 1, 535.
http://dx.doi.org/j.cirp.2013.03.017
[18] Wenig P., Kasperl S.: “Examination of the measure-ment uncertainty on dimensional measuremeasure-ments by X-ray computed tomography”, Proceedings of 9th European Conference on NDT, Berlin, Germany, September 25–29, 2006.
[19] Amirkhanov A., Froehler B., Reiter M. et al.: “Uncer-tainty in CT Metrology: Visualizations for Explo-ration and Analysis of Geometric Tolerances”, Pro-ceedings of iCT Conference, Wels, Austria, February 25–28, 2014, p. 195.
[20] Wohlers Associates Inc. Source: Material Shitnes, Stratasys, Redeyeondemand.com/Materials_FDM_ Termoplastics.aspx (access date 2011.03).
[21] Dziubek T., Pelc M.: Mechanik 2015, 12, 62. http://dx.doi.org/10.17814/mechanik.2015.12.558 [22] Budzik G., Grzelka M., Cygnar M. et al.: Technical
Ga-zette 2014, 21, No. 2, 417.
[23] Grzelka M., Jakubowicz M., Gaca M.: Mechanik 2015,
12, 91.
http://dx.doi.org/10.17814/mechanik.2015.12.563 [24] Cygnar M., Budzik G., Sendyka B. et al.: Technical
Ga-zette 2014, 21, No. 4, 867.
Rapid Communications
Przypominamy P.T. Autorom, że publikujemy artykuły typu Rapid Communications. Prace oryginalne
wyłącznie w języku angielskim o objętości 4 – 5 stron maszynopisu z podwójną interlinią (plus
ewen-tualnie 2 – 3 rysunki lub 1 – 2 tabele), którym umożliwiamy szybką ścieżkę druku (ok. 4 miesiące od chwili ich otrzymania przez Redakcję). Artykuł należy przygotować wg wytycznych zamieszczonych we wskazówkach dla Autorów.
